This classic textbook provides a unified treatment of spectral approximation for closed or bounded operators as well as for matrices. Despite significant changes and advances in the field since it was first published in 1983, the book continues to form the theoretical bedrock for any computational approach to spectral theory over matrices or linear operators. This coverage of classical results is not readily available elsewhere. It offers in-depth coverage of properties of various types of operator convergence, the spectral approximation of non-self-adjoint operators, a generalization of classical perturbation theory, and computable errors bounds and iterative refinement techniques, along with many exercises (with solutions), making it a valuable textbook for graduate students and reference manual for self-study.